Advogato stats

Posted 3 Nov 2000 at 13:46 UTC by bagder Share This

Nothing is like numbers and useless statistics. Who is the most oftenly certified Master on advogato? Who participates in most projects? Who has certified most people as Master?

I think some statistics around users and certificates could serve as input for the experiment this site is said to run.

Have a peek at the very basic main stats result for a feel.

You can have a view on the scripts that generated all that in that directory. Feel free to modify them and do better stats.

I don't claim to have made any break-throughs here, I just throw in my fascination of statistics and a few scripts that outputs lots of tables.

Enjoy!


certification distribution, posted 3 Nov 2000 at 16:31 UTC by jmelesky » (Apprentice)

Something I had noticed a few weeks ago, and was thinking about posting an article about, is the strange distribution of certifications. Observers make up 56% of Advogato members, which isn't all that unusual. But then, the Apprentice/Journeyman/Master sequence goes: 9%/27%/7% -- not exactly your traditional pyramid.

Is this valid? Are there really fewer Apprentices out there than Journeymen? Or is there something else going on? Is there a weakness in the trust mechanism? Or maybe a fundamental facet of social psychology at work?

Thoughts?

certification distribution, posted 3 Nov 2000 at 17:00 UTC by ovek » (Master)

Or simply that most apprentices don't bother to register themselves on advogato? I wouldn't expect it to be a high profile thing for them...

hm... flawed trust metric, posted 3 Nov 2000 at 17:37 UTC by ovek » (Master)

Some looking around shows that it could be the trust metric, too. I'm certified as Master by 3 people, with an obvious result. But looking around further, you can find people certified as Apprentice by 5 people, but still listed as Observer... as well as a person certified as Apprentice by 7 people, Journeyman by 1 person, and Master by 1 person, with the end result being listed as Master... I'd say the trust metric needs revision.

A sociologic factor that would compound the effect is that people are less likely to look for and certify apprentices (they'd be less likely to even know them), so apprentices would probably get less certificates. But the way the trust metric works now, it seems apprentices need more certificates than masters...

Sertification level is irrelevant, posted 3 Nov 2000 at 18:01 UTC by Zaitcev » (Master)

Only the list of certifiers is interesting.

This was brought up already when a bunch of FreeBSD goofs showed up and certified each other as master. One of them was close to the trust root and the rest is history.

Raph told you already that certification works only if majority of Advogato members apply the certification metrics as instructed. Instead, the majority uses certification as an old buddy network, with levels higher for those who are better known to them.

To get your perceptions right, use lynx. Color coded entries come up all the same in it.

Not all buddy system, posted 3 Nov 2000 at 19:29 UTC by deekayen » (Master)

While I'll agree that there is probably a bit of certification of friends, who's to say that it's still not reasonable? Like my diary points out, I've given out something like 60 certs, but only maybe 2 were people I knew and they got what they deserved.

I think the flaw is in the metric. It doesn't make sense for me to have 21 certs, and be a journeyer for just having 2 journeyer certs. I would say that maybe there should be an "average" of the certs you get, but then that wouldn't really take into account the more "trusted" certifiers vs the "untrusted" ones in the bad nodes. I don't think the diagrams shown on the trust metric docs are accurately depicted. Go to google and type in "Ford-Fulkerson algorithm" and all the university sites you'll come to will have a picture of the algorithm showing a diamond shaped picture.

ovek: I think you're wrong. I should be an apprentice and I'm not.

eh?, posted 3 Nov 2000 at 21:48 UTC by ovek » (Master)

Wrong? My first comment was uninformed, but I admitted that - but I don't see anything wrong in my second comment... you have 2 certifications for Journeyer and 19 for Apprentice, and you're still not Apprentice. Conforms perfectly with "seems Apprentices need more certificates than [higher level]", since you only needed 2 Journeyer certs out of 21 to get there... going further, you'd probably only need 1 Master cert to get to that level, too... while apprentices need a bunch of certs for their level. Sigh.

Abused Trust Part Two, posted 3 Nov 2000 at 21:53 UTC by harrisj » (Master)

For those of you who missed it the first time around, there was an interesting discussion a few months back about Abused Trust. This also inspired quite a few diary entries around the same time, but those aren't as easy to retrieve.

Zaitcev, "a bunch of FreeBSD goofs"?, posted 3 Nov 2000 at 23:00 UTC by lilo » (Master)

Pete,

You may be interested in knowing how that happened. I actually spent a fair amount of time on a BSD-related coder channel, making sure that I knew who enough of the BSD coding masters were. And spoke to a couple of people close to the trust root about making sure that known FreeBSD masters were able to certify. People near the trust root here took care of it.

If your point was that random FreeBSD enthusiasts suddenly became masters, that would turn out not to be the case. If your point is that FreeBSD is not an open source project, I would have to pass on a discussion of religion.

No offense intended, but I thought you might want to know how that came about.

lilo

Certification criteria, posted 4 Nov 2000 at 10:54 UTC by mpruett » (Observer)

Looking at the certification guidelines, there doesn't seem to be much room for interpretation in bestowing certification. The only room for critical judgement is interpretation of the terms "important," "contribute," and "significant."

The only reason to include a human element as input to this certification process is the inclusion of subjectivity. With the low level of subjectivity available in these guidelines, why should one bother certifying anyone at all when the point is just to apply these rules methodically?

If one were to assume that everyone on Advogato followed the above guidelines strictly, there would seem to be some confusion among many members regarding the meaning of these sentences:

A Master has command of the tools and is an excellent programmer.

A journeyer contributes significantly to an important free software project, or is the author of a useful or technically innovative project.

Many Advogato participants seem to realize that the guidelines aren't very participatory and have each created their own criteria. The result is that people get annoyed when others' certifications and work are not in line with Advogato's official guidelines. Rather than get frustrated with people for not following these guidelines, perhaps the answer is simply to assign no meaning other than popularity to one's certification level.

A web of trust is not well matched to the task of assessing a person's contribution to free software.

Finer granularity?, posted 4 Nov 2000 at 14:37 UTC by Michael » (Journeyer)

$gt; But then, the Apprentice/Journeyman/Master sequence goes: 9%/27%/7%

I think that perhaps another level between apprentice and journeyer is needed. Part of the problem I think comes from what makes a free software project important? So we have:

  • Apprentice: "An apprentice is someone who has contributed in some way to a free software project" and "Apprentice spends a significant amount of time learning the craft of software development"
  • Journeyer: "A journeyer contributes significantly to an important free software project, or is the author of a useful or technically innovative project."

There seems to be no place in this list for what I think most Journeyers on Advogato are: skilled coders working on largish projects, but not necessarily *important*. So for example, I consider myself above apprentice - I'm an experianced programmer and have worked on large and complex open source projects such as Nemesis and SWARM for example. However, I'd not count either of these as having current social importance. I've also worked on a reasonable sized closed source project for a company. So where do I belong? I'm above apprentice and below journeyer. From looking around I'd say that a lot of people fall into this catagory.

I also think it's useful to note that, contrary to some of the comments here, those that know me have rated me as apprentice - they both appreciate the trust metric - I'm not involved in an important project.

Apprentice also seems to cover both those who don't contribute much and those who don't have the skills to contribute much. Thus a really good coder with little time is rated equally with a first year undergrad.

Finally - who defines what an important project is?

-- Michael

Granularity and nomenclature, posted 6 Nov 2000 at 02:28 UTC by Pseudonym » (Journeyer)

Michael wrote:

There seems to be no place in this list for what I think most Journeyers on Advogato are: skilled coders working on largish projects, but not necessarily *important*.

Your proposed measure is actually recording two orthogonal things. Ideally, if this was important to you, you'd want two measures: skill/experience level (in whatever capacity, be it designing, programming, writing, public relations or what have you) and level of contribution to the community.

Part of the problems are that words like "Apprentice" and "Master" are quite misleading. I, a Journeyer, am probably more skilled and more experienced than some people who are legitimately "Masters" and less skilled/experienced than some people who are legitimately "Apprentices" according to the strict Advogato definitions. Maybe we need to use terms that refer to the concept of citizenship rather than skill.

Finally - who defines what an important project is?

I dunno. Is the GNU assembler an important project? Despite the fact that the overwhelming majority of compiled "free" or "open source" software depends on it, it's not "cool" and it's not high-profile.

granularty in the nth-dimension, posted 6 Nov 2000 at 13:15 UTC by Michael » (Journeyer)

Pseudonym wrote:

Your proposed measure is actually recording two orthogonal things. Ideally, if this was important to you, you'd want two measures: skill/experience level (in whatever capacity, be it designing, programming, writing, public relations or what have you) and level of contribution to the community.

You're right - I'm aware that I'm trying to express two things here. The problem is that I don't thing Advogato is going to get any better with multiple dimensions of ratings. To be honest I'm not sure how you could best encapsulate what I want, or even if you can produce a single simple mechanism.

Perhaps I've just got the wrong spin on the community here, but I automatically relate a person's rating to how good a coder/software engineer they are. Whereas I think the point of Advogato is more to try and develope an open source community regardless of programming ability.

If we just consider open source contribution-ness then I think we still need another level between apprentice and journeyer. Something like:

  • Apprentice: An apprentice is someone who has contributed in some way to a free software project or has their own small project (an X notepad, whizzy load meter, etc.).
  • Cook: A cook contributes significantly to a medium open source project (a mail tool perhaps, CPU simulator <grin>).
  • Chef: A chef contributes significantly to a large or several medium free software project.
  • Master: A master is a "community" guru (where community can be something like a BSD type, Linux, gcc, perl/python, KDE, Office type thing).

Feel free to change the names I suggested. I'd call myself a cook. I'm still missing out on important things here, like teamwork. Doing a project on your own is fine, but working as part of a team is also an important skill.

And, at the end of the day I imagine it's partly an ego thing - I imagine a lot of people don't want to be called an apprentice, or want to be high ranking.

And finally (just as you thought I'd ran out of stream :-):

Is the GNU assembler an important project? Despite the fact that the overwhelming majority of compiled "free" or "open source" software depends on it, it's not "cool" and it's not high-profile.

The GNU assembler should be a high profile project - I think that people should consider not only how cool a project is (coolness does come into it), but more importantly how important is a given piece of software to how large a group of people. Things like gcc, sendmail, perl, apache, cups, etc. are not as cool as say KDE, GNOME, mozilla, etc., but are fundemental to the community, and it's the community that the ratings seem to be about.

Perhaps we should throw it all away and stick out geek code at the top of our personal page. Can someone write a perl script to turn a geek code into a shade of colour? Actually, if we rated people based on open-source contributions, coding skill and helping others, then we could generate everyone a new colour with each cscale mapped to one of RGB. :-)

Alright, I'll stop ranting now :-)

--Michael

navel contemplation, posted 7 Nov 2000 at 14:25 UTC by lkcl » (Master)

okay. the way that the certs work is it's effectively "hierarchical access control lists". from the algorithms used: outcome 1) no person can certify anyone else at a level higher than they are already certified. outcome 2) those people who are designated as "seeds" are automatically at the highest level. outcome 3) the _number_ of certifications received is irrelevant: if you have one Master cert from someone who is a Master, that is enough to make _you_ a Master.

so, the distribution is simply an interesting reflection of The Way Things Are.

raph and i talked a while back about means to say, "master can only be certed as master by 5 other masters". in trust metric terms, this can be expressed as "there must be 5 independent paths from you to the seed - all at Master level", where a path is one Master to another Master to another Master etc. i finally understood what raph was describing to me :)

we seem to be going full circle again, in six months. is there this much discussion about moderation itself, on slashdot? :) :)

i remember there was issues with developer, lead developer, documenter, helper for projects. i dealt with this by turning project status / person relationship into a trust metric, splitting it into two [one for developer status, the other a documentor status]. see demo.linuxcare.com.au/proj for details. i think something like this on the live advogato.org would help clarify these issues. it will require a script or some code to modify the site database files.

luke.

Re: Advogato stats, posted 12 Nov 2000 at 23:09 UTC by okcrum » (Master)

It's possible the Apprentices are True Lurkers, even here.

Is this really an issue?, posted 14 Nov 2000 at 18:59 UTC by JoshO » (Journeyer)

Is there really a need for debate on this issue? After all the classifications are merely one members perceptions of another member based on diary entries. In my opinion everyone that is here and takes the time to make a diary entry or post an article is contributing to the community in some way, which is good, right? Perhaps the system is not accurate but even if it were, would it make a difference?

New Advogato Features

New HTML Parser: The long-awaited libxml2 based HTML parser code is live. It needs further work but already handles most markup better than the original parser.

Keep up with the latest Advogato features by reading the Advogato status blog.

If you're a C programmer with some spare time, take a look at the mod_virgule project page and help us with one of the tasks on the ToDo list!

X
Share this page